Cool Sea Surface Temperature (SST) anomalies tend to prevail in the Seychelles Dome region (Southwest Indian Ocean, NE of Madagascar) during the Northern hemisphere summer-autumn 1.5 years before El Niño. This West Indian Ocean precursor might potentially help to predict El Niño/Southern Oscillation (ENSO) events, hence we investigate its statistical significance and possible underlying mechanisms.
A specially designed test, which filters out possible influences of prior ENSO events on the Indian Ocean, shows that the correlation between the west Indian Ocean SST and ENSO at 1.5 years lead-time is indeed significant.
Further analysis of the observational record, together with result from a simplified model (an Indo-Pacific extension of the Zebiak-Cane or ZC model), suggest the following, possibly co-acting, mechanisms:
- “Atmospheric bridge”: Cool SSTs in the West Indian Ocean (WIO) lead to a local cooling and subsidence of the overlying air masses, which is compensated by a weak uplift of air away from the WIO. Above Indonesia, where a warm and moist background state prevails, this initial uplift can be amplified by convection (latent heat release invigorating the upward motion). This leads to wind convergence at the surface and thus easterlies over the Pacific. The easterlies “push” warm water towards Indonesia, creating a large warm water volume, which in turn is a prerequisite for El Niño in the following season.
- “State-dependent noise”: a cool WIO in summer-autumn is often followed after a few months by enhanced wind variability on intraseasonal timescales over the west Pacific, although the exact mechanism for this remains unclear. Such wind variability in late winter-spring, in particular westerly wind bursts, may trigger El Niño development.
However, an analysis of the output of a state-of-the-art climate model (CESM) suggests that the East Indian Ocean temperature has more influence on ENSO than the West. This may mean that the observational record is still too short or inaccurate. However, CESM results show an unrealistically strong SST variability in the East Indian Ocean. This large signal may “mask” the influence from the WIO.
The bias seems to be linked to errors in the time-mean vertical temperature profile in the East Indian Ocean, so it might be possible to reduce the bias by improving the temperature profile. This may also apply to other climate models suffering from similar biases.
We also used the ZC model to investigate the effect of long-term temperature changes in the whole Indian Ocean on ENSO. In the ZC model, a warming in the Indian Ocean leads to easterlies in the Pacific and a more La Niña-like background state. The amplitude of the ENSO cycle increases and the spatial pattern changes such that the largest temperature anomalies shift westwards from the South American coast towards the central Pacific.
The latter result is in line with observations, because in recent decades we saw both a warming of the Indian Ocean relative to the Pacific and an enhanced frequency of “central Pacific” El Niños. However, these results should be tested with more complex models than the ZC model.